'===========================================================================
' Subject: MESSAGE FILTER                    Date: 07-28-93 (00:00)       
' Author:  Victor Yiu                        Code: QB, PDS, VBDOS         
' Keys:    MESSAGE,FILTER                  Packet: TEXT.ABC
'===========================================================================
; ============ MessageFilter v1.0 =============
;
; Strips all LF (ascii 10) and ascii 141s
;
; ======== by Victor Yiu, July 25, 1993 =======
;
; v1.0 -- First release using hybrid in-line code
;
; QB users:
;    DECLARE FUNCTION StripLF% (BYVAL Segm%, BYVAL Offst%, BYVAL Leng%)
;    Text$ = LEFT$(Text$, StripLF(VARSEG(Text$), VARPTR(Text$),  LEN(Text$))
;
; PDS (far string)/VB-DOS users:
;    DECLARE FUNCTION StripLF% (BYVAL FarAddr&, BYVAL Leng%)
;    Text$ = LEFT$(Text$, StripLF(SSEGADD(Text$), LEN(Text$)))
;
; Function returns new length of string, so you should truncate old
; string with a LEFT$(Text$, ResultingCode) after calling this func.
; The reason for different syntaxes is for speed.  BTW -- this routines
; automatically adjusts segment/offset overflows, so no need to adjust them
; in QB program.
;
; This routine is fast!  It minimizes jumps and uses partial in-line coding.
;
CODE SEGMENT PARA PUBLIC 'CODE'
    ASSUME CS:CODE, DS:nothing, ES:nothing, SS:nothing
    PUBLIC StripLF

LineFeed        EQU     10
OtherBad        EQU     141

StripLF PROC FAR
        PUSH BP
        MOV BP, SP          ; set up stack frame
        PUSH DS             ; save required variables
        PUSH SI
        PUSH DI

        LES AX, DWORD PTR SS:[BP+8]     ; load starting and ending addresses
        CLD                             ; go forward

        MOV DI, AX          ; save copy of offset in DI
        MOV CL, 4           ; divide offset by 16 to add to segment
        SHR AX, CL
        MOV DX, ES          ; add to segment
        ADD DX, AX
        AND DI, 15          ; adjust offset after operation
        MOV ES, DX
        MOV BX, DI          ; save offset in BX for later
        MOV DS, DX
        MOV SI, DI

        MOV CX, WORD PTR SS:[BP+6]      ; get count
        MOV AX, CX              ; get remainder in AX
        SHR CX, 1               ; divide loop count by 8
        SHR CX, 1
        SHR CX, 1
        AND AX, 7               ; mask
        MOV DL, 10              ; prepare to x10 AL
        INC CX                  ; to take into account the extra loop
        MUL DL
        NEG AX
        ADD AX, OFFSET LoopCheck
        MOV DX, (LineFeed * 256) + OtherBad     ; preload for speed
              ;     DH                DL
        JMP AX

EVEN            ; each byte processed is 10 bytes of code
StripTop:
        LODSB               ; load byte
        CMP AL, DH          ; LF?
        JE Byte2            ; yep -- skip
        CMP AL, DL          ; 141?
        JE Byte2            ; yep -- skip
        STOSB               ; just store then...
Byte2:  LODSB
        CMP AL, DH
        JE Byte3
        CMP AL, DL
        JE Byte3
        STOSB
Byte3:  LODSB
        CMP AL, DH
        JE Byte4
        CMP AL, DL
        JE Byte4
        STOSB
Byte4:  LODSB
        CMP AL, DH
        JE Byte5
        CMP AL, DL
        JE Byte5
        STOSB
Byte5:  LODSB
        CMP AL, DH
        JE Byte6
        CMP AL, DL
        JE Byte6
        STOSB
Byte6:  LODSB
        CMP AL, DH
        JE Byte7
        CMP AL, DL
        JE Byte7
        STOSB
Byte7:  LODSB
        CMP AL, DH
        JE Byte8
        CMP AL, DL
        JE Byte8
        STOSB
Byte8:  LODSB
        CMP AL, DH
        JE LoopCheck
        CMP AL, DL
        JE LoopCheck
        STOSB
LoopCheck:
        LOOP StripTop

        MOV DX, DI      ; get new offset count
        SUB DI, BX      ; get length
        MOV AX, DI      ; return result
        POP DI
        POP SI          ; restore registers
        POP DS
        POP BP
        RET 6           ; shave off 6 bytes -- new length returned in AX
StripLF ENDP
End

;============== end of asm. code.  Look at next message for BASIC
; program to test it.

'Here's the BASIC source to utilitze the asm. routine? ---------
' ======== CFilter 1.0 =======
' asm. by Victor Yiu
'
' QB users:
'    DECLARE FUNCTION StripLF% (BYVAL Segm%, BYVAL Offst%, BYVAL Leng%)
'    Text$ = LEFT$(Text$, StripLF(VARSEG(Text$), VARPTR(Text$), LEN(Text$))
'
' PDS (far string)/VB-DOS users:
'    DECLARE FUNCTION StripLF% (BYVAL FarAddr&, BYVAL Leng%)
'    Text$ = LEFT$(Text$, StripLF(SSEGADD(Text$), LEN(Text$)))
'
' Function returns new length of string, so you should truncate old
' string with a LEFT$(Text$, ResultingCode) after calling this func.
' The reason for different syntaxes is for speed.  BTW -- this routines
' automatically adjusts segment/offset overflows, so no need to adjust them
' in QB program.

DECLARE FUNCTION Filter$ (Text$, Array() AS STRING)      'one you call
DECLARE FUNCTION StripLF% (BYVAL FarAddr&, BYVAL Leng%)  'asm func.
DEFINT A-Z

OPEN "Sample" FOR BINARY AS #1
Text$ = INPUT$(LOF(1), 1): PRINT "Size:"; LOF(1)
CLOSE : IF LEN(Text$) = 0 THEN KILL "sample": PRINT "File?": END
CLS

DIM Array() AS STRING

       ' Text$ is modified -- LFs and 141s are stripped
Address$ = Filter$(Text$, Array())
Text$ = ""

OPEN "Output" FOR OUTPUT AS #1
FOR Num = 1 TO UBOUND(Array)
    PRINT #1, Array(Num);
NEXT
CLOSE

PRINT Address$

FUNCTION Filter$ (Text$, MsgText() AS STRING)
    REDIM MsgText(1 TO 200) AS STRING
    AP = 1: CR$ = CHR$(13)

    ' ==== the FAST asm. procedure
    Text$ = LEFT$(Text$, StripLF(SSEGADD(Text$), LEN(Text$)))

    Prev = INSTR(Text$, CR$ + CHR$(1) + "MSGID: ")
    IF Prev THEN SpaceAfter = INSTR(Prev + 8, Text$, " ")
    IF SpaceAfter THEN
        Address$ = MID$(Text$, Prev + 8, SpaceAfter - Prev - 7)
    END IF
    IF RIGHT$(Text$, 1) <> CR$ THEN Text$ = Text$ + CR$

    Pntr = 1: Prev = 1: CurrentTop = 200
    DO
        Pntr = INSTR(Pntr, Text$, CR$)
        IF Pntr THEN                    ' dont include CR$ after each line
            MsgText(AP) = MID$(Text$, Prev, Pntr - Prev)
            Pntr = Pntr + 1: Prev = Pntr: AP = AP + 1
            IF AP > CurrentTop THEN
                CurrentTop = CurrentTop + 50
                REDIM PRESERVE MsgText(1 TO CurrentTop) AS STRING
            END IF
        END IF
    LOOP WHILE Pntr
    AP = AP - 1
    REDIM PRESERVE MsgText(1 TO AP) AS STRING

    '
    'This portion of code looks for ^aMSGIDs and parses them.  If it doesn't
    'find a valid Fidonet address in the MSGID, it looks for the origin line
    'address (Inside brackets)
    '
    'This portion of code follows directly after the above code.

    IF INSTR(Address$, ":") = 0 OR INSTR(Address$, "/") = 0 THEN
        FOR X = AP TO 1 STEP -1
            IF LEFT$(MsgText(X), 11) = " ! Origin: " THEN
                LG = LEN(MsgText(X))
                IF ASC(RIGHT$(MsgText(X), 1)) = 41 THEN ')

                    FOR Y = LG TO 11 STEP -1
                        IF ASC(MID$(MsgText(X), Y)) = 40 THEN
                            EXIT FOR        '(
                        END IF
                    NEXT
                    Address$ = MID$(MsgText(X), Y + 1, LG - Y - 1)
                END IF
            END IF
        NEXT
    END IF

    Filter$ = Address$
END FUNCTION